Edit detail for StructureCalculationWithAtnosCandid revision 11 of 16

Structure calculation with AtnosCandid

Navigation: return to StructureCalculation? general page.

Guidelines for structure calculation using Unio together with CARA.

1. Make sure all your spectra are correctly calibrated from the beginning.
   • It is far easier to correctly calibrate your spectra at the beginning of a project than to fix incorrect calibrations at a later stage of the project. Inconsistent calibrations between different spectra can cause lots of extra work. Incorrect absolute chemical shift calibrations can result in the misassignment. CARA relies on chemical shift statistics to guide the assignment process!
   • Therefore always reference your spectra to DSS. Once you have decided on your NMR conditions for assignment and structure determination, prepare an NMR sample with your protein (or just your NMR buffer) and add a small amount of DSS to it.
   • At the beginning of each NMR session calibrate the temperature and measure a high resolution 1D of this reference sample and process it with zerofilling (e.g. 32k). Directly reference the proton shifts by setting the DSS signal to zero ppm. In Brukers Topspin this will modify the value of the processing parameter SR (the difference in Hz between the frequency of the DSS resonance in the sample and the 1H basis frequency BF1 of the spectrometer).
   • indirectly reference all spectra measured with the protein using this 1D reference calibration.
   • For carbon and nitrogen shifts use the ratios of the heteronuclear to the proton gyromagnetic ratios: See BMRB.
   • Be aware that the ppm scale of spectra processed by Topspin currently has a one half pixel shift in the position of peak maxima. I.e. the same spectrum processed with different zero-filling in a given dimension will show a systematic shift in the position of the peak maxima in that dimension. This is true both in Topspins spectrum display and for the same spectra displayed in CARA. To avoid this you can use the AU script DSSref (a modified version of the AU script calc_wishart.dmo included with the bruker Topspin distribution). This script uses the above Chi(Xnucleus)/Chi(1H) ratios of gyromagnetic ratios to indirectly reference heteronuclear dimensions of ND experiments and also corrects all dimensions ppm scales for the one half pixel offset. See the header of the script for details on usage.
   • When a new spectrum is added to the project, take time to ensure that it is calibrated to match well to all other spectra in the project. For 3D spectra it is helpful to calculate the 2D projections, and calibrate these projection spectra. Then the calibration values (cal in the SpectrumExplorer?) can be transferred to the corresponding dimensions of the 3D spectrum. E.g. HNCA. There are two 2D projections: 2D HN(CA) which can be loaded with SpectrumType? 2D HSQC15N and 2D H(N)CA which can be loaded with SpectrumType? 2D HnCA
2. Complete the sequence-specific resonance assignments of the backbone and sidechains.
   • Check your assignments using the LUA script AssignmentReport.lua. This will help you identify the following types of inconsistencies before starting structure calculation:
     • Unusual shifts. "Outliers". These shifts occur 4 standard deviations outside of the mean. Did you accidentally swap the assignments of neighboring CH2 groups? This can happen if you used only TOCSY-type spectra to support sidechain assignment.
     • Incompatible labels. Stereopairs should not be assigned if you made a pseudoatom assignment. It does not make sense to have both HB2 and HB assigned in an AMX system like SER. Either delete the HB2 & HB3 leaving only HB or delete HB keeping only HB2 and HB3. Note that a stereospecific assignment is indicated by an exclamation point in front of the label: !HB2.
     • Missing assignments. Unio requires 90% of BB and 85% of sidechain assignments. AssignmentReport gives you overall statistics as well as a list of missing assignments reported for each residue.
3. Write out the calibrations from the CARA project to the spectra. Right-Click on each NOESY spectrum in the Spectrum-explorer and select "Write Calibration".
   • Note that this option is currently only available for XEASY spectra. If your NOESY spectra are in another format, you will first have to convert them to XEASY format to use this command.
     • To do this, you can use the CARA main menu "Tools-Convert to EASY spectrum".
     • Then click on each NOESY spectrum in Spectrum Explorer and right-click "Replace Spectrum" selecting the XEASY spectrum .param file
     • Then "Write Calibration" for each NOESY spectrum. (To learn why this is necessary: click here)
4. Adjust the chemical shifts (spin positions) so that they match well to the NOESY spectra you will use for structure calculation.
   • If you are only working with 2D NOESY spectra, view the NOESYs in HomoScope? or PolyScope? and adjust the positions of the spins so that the signals observed in the NOESY towers are centered well to the peak positions. Click on a peak in the system and move the cursor with arrow keys to center on the NOESY tower and then type the "move spins" command "ms".
   • If you are working with 3D NOESY spectra, open the 2D projection of that 3D spectrum (H-N for 15N-resolved NOESY, H-C for 13C-resolved NOESY) with PolyScope? and check that the spin systems are centered on the observed signals. It is best to do this as follows:
     1. Select the corresponding 3D NOESY spectrum in the Strips-Select Spectrum" menu
     2. Select a system either by clicking on it in the projection (2D plane) or by typing the goto residue command "gr" and typing the residue number.
     3. The NOESY tower will appear in the two strips on the right. Use the arrow keys to move the system position in the 2D projection so that the NOESY tower is centered well on the grey line in both strip views XZ and YZ.
     4. When the position matches, type the "move spins" command "ms".
   • if you have slight differences in the chemical shifts between the NOESY spectra you will use for structure calculation, it may be necessary to create alias shifts for those shifts.

     Note that if you have many large systematic shift differences between NOESY spectra, this indicates that the solution conditions or proteins may differ. It is probably better to make a fresh sample or use the same sample for all NOESY measurements to ensure that all are taken under identical conditions than to try to analyse inconsistent data.

   • To create an alias shift, click on a peak in the NOESY tower which is not centered in one NOESY spectrum (but agrees in all the other NOESY spectra) and use the arrow keys to move the cursor so that it is centered on the NOESY tower. Right-click and select "move alias".
   • If you have systematic differences in the chemical shifts of some classes of spins relative to others you can use ShiftSpinsInCatagory.lua to adjust them.
   • If you must calibrate (or recalibrate) spectra after the assignment process is completed use the script ShiftSpinsInCatagory.lua to shift spins in a given class as a group to match the position of signals in the calibrated spectra.
     1. The NOESY spectra used for structure calculation should be calibrated to DSS externally. I.e. if you use Topspin to process the NOESYs? these spectra should have chemical shift scales which are properly referenced to DSS. When you load them into the CARA project with the assignments, the NOESY tower positions should agree with the positions of crosses CARA shows for these towers. If this is not the case, they should not be recalibrated inside CARA. Instead, the chemical shifts of the project should be adjusted so the crosses generated by CARA agree with the position of NOESY signals. Do this as follows:
     2. Determine the offset in position between actual signal position and cross in each dimension of the NOESY. (e.g. in the direct acquisition dimension of a 3D 15N NOESY the cross positions are systematically 0.1 ppm higher than the real signal position).
     3. Apply this correction to all 1H atoms using the script ShiftSpinsInCatagory.lua
     4. Afterwards run the script RecalibrateSpectra to adjust the calibration values of all spectra in the project.
     5. Finally reset the cal values of the NOESY spectra used for the structure calculation to zero using the calibrate spectrum context menu in CARA's SpectraExplorer?.
     6. You will need to do this for each type of shift appearing in the NOESY spectra. (1H, 13C, 15N).
5. Generate .seq file and .prot files using the script WriteAssignments.lua.
   • Unio accepts the .seq and .prot files (XEASY format) as input.
   • Run the script: WriteAssignments.lua (Why use this script and not the menu item ""project-export atomlist"? See FAQ)
     1. select the project: which contains your final assignments and properly referenced NOESY spectra.
     2. select spectrum: If there are no significant differences in the shifts of corresponding spins from different NOESY spectra, you can leave the default setting (no aliases). However, if you see differences in the positions, and you created alias shifts for individual NOESY spectra (see above), then you can write out one proton list for each NOESY used in the structure calculation. You will need to enter each of the proton lists together with the NOESY spectrum files location in the AtnosCandidSetup file. See Unio setup page for documentation.
     3. select the library format: DYANA, CYANA1.0, or CYANA2.0+. Note that the cyana.lib included with the current distribution of CYANA is compatible with the "CYANA2.0" setting of WriteAssignments.lua.
     4. select oxidized or reduced for the CYS residues: This can be determined from the CB shift. >35ppm is oxidized. WriteAssignments.lua only allows you to select all reduced or all oxidized. If you have a mixture, you will need to edit the .seq file after writing it out to reflect the correct redox states for each CYS.
     5. determine the output mode: Terminal or File.
        • Terminal: this is just for testing the output and does not write out a file
        • File: this writes out .seq and .prot files to the directory where you started CARA from. It will overwrite any existing files with the same name.
6. Copy the .seq and .prot files to your AtnosCandid structure calculation directory. (please see Unio Setup Page for a description on how to setup and start the structure calculation)
   • Important! Make sure that your library formats in the AtnosCandidSetup.txt file match those that you selected when you wrote out the .seq and .prot files with WriteAssignments.lua.
7. Check to be sure the covalent check of AtnosCandid gives a sufficiently good chemical shift agreement between the input chemical shift list (.prot) and the corrresponding NOESY spectra. Typical values are 80-90%. If your values are lower, it may mean there is disagreement between the chemical shifts in your project and the position of the corresponding resonances in the NOESY spectra.
   • Be sure that your calibrations were written to the spectrum file which you supplied to AtnosCandid. I.e. the NOESY spectra you supplied to AtnosCandid are loaded into the project without any calibration being applied inside CARA (cal values in Spectrum-Viewer are all zero): the position of the NOESY towers in the CARA project should agree with the crosses CARA displays.
   • If this is not the case, try to improve the position of your NOE towers. Your chemical shift assignments should agree well with the NOESY spectra used in the structure calculation.
   • Possibly there are slight differences in the chemical shifts from the different NOESY spectra? You may have to adapt some chemical shift positions to the individual NOESY spectra using aliases. Then you will need to write out the shifts for each NOESY spectrum separately using the Write aliases from spectrum option in WriteAssignments.lua.
   • To evalulate the results of the covalent check and identify resonances with problems, see here.
8. Start the structure calculation. See the Unio [Tutorials]? for details.